[MoE] Generalize masked activation for padded layouts (#51217)

Misha Goin committed about 2 months ago

[Bugfix] Gate sm_100-only kernel tests on the capability family, not >= (#54306)

Bojiang Li committed about 2 months ago

[Mypy] Fix typing for M models (#54262)

Taneem Ibrahim committed about 2 months ago

[Bugfix] Reject tokenless chat and audio streams (#54708)

Taneem Ibrahim committed about 2 months ago

[Bugfix] Support MCP SDK 2.x tool input schemas (#53870)

Chauncey committed about 2 months ago

[Kernel] Add B12X causal paged attention backend (#52017)

Luke Alonso committed about 2 months ago

[Mamba] Add FlashInfer ReplaySSM backend (#52506)

Andrii Skliar committed about 2 months ago

[Attention] Enable adaptive verification for FLASHINFER_MLA_SPARSE_DSV4 (#52724)

Ilya Markov committed about 2 months ago

[Feat][MM Hashing] include media_io_kwargs in multi-modal hashes (#54241)

Jan Kwieciński committed about 2 months ago

[Bugfix][Model] Fix GraniteMoeHybrid per-expert quantized weight loading (#54052)

Priyansh Jain committed about 2 months ago

[Doc] Document FP8 GEMM kernel selection and Blackwell support (#49936)

Harjoth Khara committed about 2 months ago

[Kernel][Qwen] Add Hopper LL-GEMM tuning table for Qwen4Exp (#54560)

Zheng Cai committed about 2 months ago

[Feature][Spec] Support disabling trailing prefix-cache block dropping (#53388)

Ziming Huang committed about 2 months ago

[Bugfix] Account for client queue time in serve benchmarks (#54136)

maithilijoshi20 committed about 2 months ago

[Bugifx][INC] Fix INC quantization method selection for non-quantized layers (#47237)

Liang Lv committed about 2 months ago

[CI/Build] Fix entrypoints coverage (#54750)

Cyrus Leung committed about 2 months ago

[CI] Add repository-local OTel tracing helpers (#52851)

Kevin H. Luu committed about 2 months ago

[CPU][Zen] Route Int8 MoE inference through zentorch on AMD (#44834)

Ganesh R committed about 2 months ago

[Bugfix][KV Offload] Isolate tiering shutdown failures (#52290)

AlexHuang committed about 2 months ago

[Bugfix][MiniCPM-V] Route video_embeds to the shared vision parser (#54633)

Subhash Polisetti committed about 2 months ago